Tutorial 6: Social Media Mining in Learning Contexts

Bodong Chen
#LASI2018, June 11-13, NYC

About Me

Assistant Professor in learning technologies
University of Minnesota

  • Collaborative learning
  • Learning analytics
  • MOOCs

Acknowledgements

I do not own the copyright of any of the images in this presentation. I therefore acknowledge the original copyright. Other content in this presentation (e.g., text and code) is licensed under a Attribution 4.0 International (CC BY 4.0) license.

Disclaimer

Work included in this presentation is produced for teaching purposes only and is hereby not intended for generalizable scholarly knowledge.

Housekeeping

Feel free to ask questions ANYTIME!

Prep

Install

  1. R
  2. RStudio development environment
  3. R packages: tidyverse, igraph, sna, …

Pair-up

Tentative Agenda

  1. Social media basics
  2. Social media for learning
    (Short Break)
  3. Collecting social media data
  4. Mining social media data

Part 1: Social Media Basics

What are social media?

  1. Pair & Share
    (self-intro)
  2. Tweet out your definition(s) – #LASI18smm

What do social media researchers say?

Gruzd, Staves, & Wilk 2012:

What do social media researchers say?

Obar & Wildman, 2015:

  1. Social media services are (currently) Web 2.0 Internet-based applications
  2. User-generated content is the lifeblood of social media
  3. Individuals and groups create user-specific profiles for a site or app designed and maintained by a social media service
  4. Social media services facilitate the development of social networks online by connecting a profile with those of other individuals and/or groups

Example: Twitter

Example: Twitter

(Goggins & Petakovic, 2014)

Example: Github

Example: Github

(Goggins & Petakovic, 2014)

Example: Wiki

Part 2: Social Media for Learning

AND
Learning on Social Media

Educational Affordances

  • connectivity (interconnections)
  • interactivity
  • information sharing and discovery
  • content creation, remixing
  • etc.

(Bates; Greenhow; Siemens & Downes)

Cases

  1. Pair & Share
    • What educational use of social media you're involved/interested in?
    • How is learning conceptualized in the context?

Learning attributes on social media

Category Formal attributes Informal attributes Social media attributes
Purpose Externally determined Self-determined community of interest Self-determined socially determined
Process of Learning Teacher-led (didactic) Self-directed (negotiated); Individual agency Self-directed; Other- influenced; Unintended network effects
Location/context Time-restricted Open-ended Open-ended
Content Knowledge acquisition; High status knowledge Everyday practice; Status of knowledge unacknowledged User-generated, Re-mixed; Social construction and distribution, Knowledge as collective agreement

(Adapted from Greenhow & Lewin, 2016)

Broadening the Conceptualization & Scope

Learner-centered themes

  • Learners' participation and creative practices
  • Learners' online identity formation

(Greenhow et al., 2009)

Influence, Participation and Affiliation

(Goggins & Petakovic, 2014)

Social (and Socialized) Learning

(Buckingham Shum & Ferguson, 2012)

Context Matters!

Example: CCK08 cMOOC, Siemens

Example: Open Learning MOOC

http://openlearninghub.net/ (Garnder Campbell)

Example: Open Learning MOOC

Example: Connected Learning

Example: Connected Learning

Example: Personal Learning Networks

#personalizedpd

Example: #MarginalSyllabus

“The Marginal Syllabus convenes and sustains conversations with educators about issues of equity in teaching, learning, and education.”

Example: Scholarly Communication

Open Discussion

Part 3: Collecting Social Media Data

Privacy and Ethics

  • Research
  • Human subjects or web documents
  • Private information
  • Informed consent
  • Data, algorithms, and practices

Case Study 1: #personalizedpd

Case 1: #personalizedpd

Blog posts

At first Twitter allowed me to fill the personalized PD void and connected me with others who both challenged my thinking and inspired me to be a better educator than I was. …

Then, one glorious day, I found the #personalizedPD chat and my entire perspective changed. Instantly, I loved these people.

Mandy Froehlich

Collect Tweets

TAGS

Collect Tweets

# Load helpful functions written by Bodong
source("utils/collect_tweets.R")
source("utils/munge_tweets.R")

# URL to the published sheet
gsheet_url <- "https://docs.google.com/spreadsheets/d/e/2PACX-1vRzuK4ltTt6jSJXYAhj-ERxvlhw3UsP5zCeCyrfXfEwNyTRu8kn_Hn8RwVewaxMPly2KcJAiciCTK_0/pub?gid=400689247&single=true&output=csv"

# Collect tweets archived in the sheet
tweets_raw <- collect_tweets_from_gdrive(gsheet_url)

Clean and store data

# basic data cleanup (incl. removing duplicates)
tweets <- preprocess_tweets(tweets_raw)

# save cleaned data as a csv file
write_csv(tweets, path = "data/tweets.csv")

Case 2: #MarginalSyllabus

API keys

Collect Hypothesis Annotations

API (aka. Application Programming Interface)

# Load helpful functions written by Bodong
source("utils/collect_annotations.R")

# Collect annotations
annotations <- collect_annotations(tag = "marginalsyllabus",
                                   token = h_token,
                                   num = 500)

Store data

# save data as an Rdata file
save(annotations, file="data/annotations.Rdata")

Storing data

Porpular formats

  • CSV (comma-separated vectors)
  • JSON
  • Relational databses (e.g., MySQL)
  • GraphDB (e.g., Neo4j)

Neo4j

Troubleshooting?

Part 4: Mining Social Media Data

Mining what?

Connectivity of a community

Influence and leadership

Community evolution

Cognitive and affective states

Thank You!